IRC logs for #automotive for Friday, 2016-07-15

*** tjamison has left #automotive00:27
*** kooltux__ has quit IRC00:33
*** mdurnev has joined #automotive01:01
*** AlisonChaiken has joined #automotive05:17
*** CTtpollard has joined #automotive06:59
*** warter has joined #automotive07:28
*** warter is now known as jbocklage12307:29
*** rstreif has joined #automotive07:33
*** rajm has joined #automotive07:39
*** leon has joined #automotive07:57
*** leon is now known as leon-anavi07:57
*** jonathanmaw has joined #automotive08:00
*** kbroulik has joined #automotive08:04
*** rdale has joined #automotive08:08
*** rstreif_ has joined #automotive08:10
*** rstreif has quit IRC08:11
*** gunnarx has joined #automotive08:16
CTtpollardgunnarx: I'm debating a branch of g-d-p that tracks the current yocto/ivi version branch instead of refs and dropping it into a pipeline08:33
* gunnarx is reading, trying to understand08:34
gunnarxcan we break it down.  what is current yocto/ivi version branch?   meta-ivi?08:35
CTtpollardnot sure I can simplify that one further08:35
CTtpollardwe're defined by an ivi release, then in turn a yocto release08:35
CTtpollardwe use .gitmodules to point to refs, you can point to branches08:35
gunnarxyes, a "meta-ivi" release if you want to be more clear08:35
gunnarxHow do you point to branches?  Not using submodules I think08:36
gunnarxModifying the init script to checkout the tip of the branch?08:37
CTtpollardit can be given a branch field08:37
gunnarxit being a submodule?08:37
CTtpollardyes, since git 1.8.2 apparently08:39
CTtpollardfor meta-oic as an example08:39
gunnarxsorry I'm slow today... despite Goa Psy-Trance in the headphones. :)08:39
CTtpollardI'd blame friday08:39
gunnarxSounds good to me.  There are many ways, you could of course modify init script also.08:40
gunnarxKeeping this on a branch makes sense though - easy to set up such a pipeline.08:40
gunnarxDo we like that idea better than the pipeline idea with an environment variable that controls which layers to replace (the one that would generalize our current latest-meta-genivi-dev recipe)08:41
pedroalvarezgunnarx's pipeline idea will allow us to forget about this special branch that would track the tip of meta-ivi branch08:43
gunnarx:( they hate my idea ...08:43
CTtpollardjust one thing, does it still have the limitation of only being able to track PR's for 'master'?08:43
pedroalvarezCTtpollard: no, the limitation is that the repo has to have a master branch08:43
CTtpollardah ok08:44
gunnarxFor each individual repo, only PRs aimed to the master branch of that repo will get picked up08:44
gunnarxso master branch of m-g-d can be tracked, or any other repo.08:44
CTtpollardhmm, so for instance now with meta-ivi10 we could not use it to test PR's into gdp08:44
CTtpollardbut we could just for standard commits to that version branch08:45
gunnarxI don't see any separate meta-ivi-10 branch on g-d-p git?  what are you referring to?08:45
* pedroalvarez continues at being on holidays08:46
CTtpollardgunnarx: meta-ivi's own 10.0 branch08:46
gunnarxwhat the hell, some cheesy lovesong is invading my psychedelic trance.  hang on.08:46
CTtpollardwrong playlist? :P08:47
gunnarxok, changed channel now.  you were saying?08:47
gunnarxYes, the bug (but they probably consider it a feature request, ugh) is open.   PRs to meta-ivi branch 10 can't be automatically tested.08:48
gunnarxBut you could script that instead then.08:48
CTtpollardso, we could use the layer replacement to track a version branch of meta-ivi (say 10.0 for now) which would test each new commit against master GDP. But dues to the master limitation with PR's we could not test PR's to that branch against master gdp08:49
gunnarxI suppose it's possible to script GitHub PR monitoring separately08:50
gunnarxOne of you could go give a +1 on the bug/feature request :)08:51
gunnarxs/One/All/ !08:51
CTtpollarddefinitely, got the url?08:51
gunnarxyes, just have to go to Confluence, it's on our Go pages08:52
khemCTtpollard: it smells like you guys are working towards a CI pipeline of some sort ?08:56
CTtpollardkhem: indeed :)
khemok. git submodules or gerrit ?08:59
khemah logins08:59
CTtpollardGDP currently uses git submodules to init the needed yocto layers, is the CI system09:00
CTtpollardkhem: you can login is guest if you wish to take a look09:00
CTtpollardgunnarx: speaking of which, how does one recover/reset a password on go.genivi?09:03
gunnarxsorry, was afk a while.  recover/reset a password must currently follow the same process as creating a new account I'm afraid.09:08
gunnarxIn other words, notify me and I'll delete the old user and confirm the SSH key is in place and then you continue from there.09:08
*** jbocklage123 has quit IRC09:14
gunnarxCTtpollard:  I imagine it should be somewhat trivial to script a GitHub PR trigger...  You know you can pull down pull requests right?  And their numbers are consecutive.  So just iterate from a cached value up until the pull fails.  That's the most recent PR.  Then build it :)09:22
gunnarxYou'd need a trigger of course, but we might use a timer maybe...09:23
CTtpollardyeh without digging into the API a timer for the poll rate is probably the place to start09:25
gunnarxI must have found the right playlist now - Progressive-Psy.09:26
gunnarxbrain kicking into gear...09:27
CTtpollardI think the replacement of the core layers for now, tracking the current version/release in use is the best place to start. From there we can look into we use the feedback from there to keep gdp master up-to date09:30
gunnarxMakes sense.  If you end up still contemplating PR builds you may want to check out TOOL-86 also for info.  I can't add watchers on tickets - maybe something toscalix can fix?09:31
CTtpollardi.e new commit in say poky jethro branch is tested, green light. What mechanism will we follow to take that result and update GDP master to that submodule commit09:31
gunnarxIdk GDP maintainers do what they're supposed to ;)09:32
gunnarxEh, let me back up09:32
gunnarxSure, automating a new PR possibly.  Not sure if it's the best use of our time, but why not set up a "wish list" ticket on it.09:32
CTtpollardThere needs to be some sort of feedback loop from that sort of testing, else it's pretty noisy09:33
gunnarxwhat do you have in mind?09:34
CTtpollardnot quite sure yet, but generating a matching PR sounds like an ultimate goal. But even some sort of notification with a parse-able such a this sha of gdp was used with this sha of $metalayerbranch succeeded09:37
CTtpollard*output such as09:37
gunnarxYes.  I was going to end it with my findings but I put the JIRA ticket back to in progress to capture this stuff also.09:38
CTtpollardfor now, unless you know of something else, we'd have to just trawl through the job logs from the pipelines to see what commits were tested09:38
CTtpollardI'll dump my feelings into the ticket, thanks for pointing it out09:39
gunnarxI know what you're after and it is useful.  But many times you are OK with knowing that the topmost commit works.  E.g. if some branch was failing but now works, you'd update the layer to what now works and ignore the previous.09:39
gunnarxSo in practice it usually works out, but definitely if automation can tell you every combination...09:39
gunnarxautomating git bisect and such is a holy grail.09:40
CTtpollardyes I suppose if the last job is green in such a pipeline, then that infers the tip of the branch is ok to be integrated09:43
CTtpollardwhich could be assessed at a given rate, daily / weekly perhaps09:45
gunnarxYes.  The Go/GitHub PR builds work that way - they only test the latest commit in the PR.  If you add a new commit, the test is rerun.  (I think you know this but for everyone's benefit...)09:46
CTtpollardthen the manual PR to switch GDP to any of the new tested sha's from the modules will allow the changes to be tested in a single integration09:48
CTtpollardas that could potentially lead to an error09:50
gunnarxisn't that the case today?  If you sent a PR to g-d-p.git you can include multiple layers and submodule updates if they have dependencies?09:52
CTtpollardyes, but what I mean is that we could have these pipelines tracking the tips of version branches, with will be integrated against master of GDP09:53
CTtpollardsay if we did a weekly check, and 'x' layers were tested and shown to have new working commits, they still need to be tested in a single integration09:54
CTtpollardsorry if I'm not getting the point across very well09:57
gunnarxNo it's fine, I see what you mean.  But it feels like we're beating a dead horse by now - why not implement something and try it for a while.09:57
CTtpollardit doesn't help that im also trying to handle my actual car's problems at the same time as this discussion09:58
*** gunnarx has quit IRC10:21
*** gunnarx has joined #automotive10:26
*** gunnarx has joined #automotive10:26
*** nuohan has quit IRC11:55
*** nuohan has joined #automotive11:59
*** nuohan has quit IRC12:09
*** rstreif has joined #automotive12:09
*** rstreif_ has quit IRC12:10
*** nuohan has joined #automotive12:11
*** jonathanmaw has quit IRC12:16
*** jonathanmaw has joined #automotive12:31
*** rstreif has quit IRC12:39
*** rstreif has joined #automotive12:39
*** rstreif_ has joined #automotive12:52
*** rstreif has quit IRC12:52
*** rstreif_ has quit IRC12:58
*** rstreif_ has joined #automotive12:58
*** ashwasimha_ has joined #automotive13:06
gunnarxmobile networks. it's the same for me a lot of time... :-/13:13
*** rstreif_ has quit IRC13:42
*** mdurnev has quit IRC13:45
*** jlrmagnus has joined #automotive13:49
*** nuohan has quit IRC13:52
*** halfline has quit IRC14:03
*** halfline has joined #automotive14:04
*** nuohan has joined #automotive14:10
*** nisha has quit IRC14:14
*** jlrmagnus has quit IRC14:26
*** nuohan has quit IRC14:41
*** nuohan has joined #automotive14:41
*** nuohan has quit IRC14:47
*** nuohan has joined #automotive14:47
*** nish has joined #automotive14:51
*** nish is now known as nisha14:51
*** kbroulik has quit IRC15:06
*** gunnarx has quit IRC15:08
*** jlrmagnus has joined #automotive15:37
*** jonathanmaw has quit IRC15:45
*** mvick has quit IRC15:52
*** gunnarx has joined #automotive15:57
*** gunnarx has joined #automotive15:57
*** mvick has joined #automotive15:59
*** ashwasimha_ has quit IRC16:11
*** leon-anavi has quit IRC16:15
*** rajm has quit IRC16:28
*** gunnarx has quit IRC16:30
*** rdale has quit IRC16:44
*** AlisonChaiken has quit IRC16:47
*** jlrmagnus has quit IRC17:11
*** AlisonChaiken has joined #automotive17:17
*** robin__ has quit IRC17:44
*** jlrmagnus has joined #automotive17:52
*** jlrmagnus has quit IRC17:57
*** robin__ has joined #automotive18:22
*** CTtpollard has quit IRC18:30
*** praneeth has quit IRC19:31
*** praneeth has joined #automotive19:31
*** Martiini has joined #automotive19:41
*** praneeth has quit IRC19:54
*** jlrmagnus has joined #automotive20:12
*** praneeth has joined #automotive20:17
*** jlrmagnus has quit IRC20:29
*** khem has quit IRC20:33
*** jlrmagnus has joined #automotive20:44
*** khem has joined #automotive20:45
*** nisha has quit IRC21:02
*** murphy has quit IRC21:17
*** jlrmagnus has quit IRC21:47
*** jlrmagnus has joined #automotive21:47
*** Martiini has quit IRC22:51
*** mvick has quit IRC22:59
*** jlrmagnus has quit IRC23:58

Generated by 2.14.0 by Marius Gedminas - find it at!